<![CDATA[The research discusses the simulation of a GPS-based attendance coordinate point authenticity verification system. Verification is carried out using the Isolation Forest model to detect outliers based on distance anomalies between entry and exit attendances and total path anomalies combined with Threshold-based Filtering to determine the normal distance threshold, and LSTM to analyze temporal patterns based on the total recorded path. The test results show that the combination of Threshold-based Filtering, Isolation Forest, and Long Short-Term Memory (LSTM) is able to detect invalid coordinate points accurately, from this combination obtained accuracy results of 99.74%, precision 99,49%, recall 100% and F1-score 99,74%. These results prove that the performance of the model combination (hybrid) is superior to using each component model separately.]]>
Copyrights © 2026
